DEGandMore:

title: "RNA-seq 2G" author: "Zhe Zhang" date: "2016-09-26" output: html_document: number_sections: yes self_contained: no toc: yes toc_float: collapsed: no

**RNA-seq 2G** is a web portal with >20 statistical methods that perform two-group analysis of differential gene expression. It uses read count data from RNA-seq or similar data matrix as input and generates test statistics in consistent format as output.

Introduction

Two-group comparison of differential expression (DE) is the most common analysis of transcriptome data. For RNA-seq data, the comparison is usually performed on a gene-level matrix of read counts, with the read counts corresponding to the number of sequencing reads mapped to each gene in each RNA-seq sample.

Statistical methods that have been applied to two-group DE of RNA-seq data are widely different in terms of their data distribution assumption, input/output format, performance, sensitivity, and user-friendliness.

**Table 1** DE method features.

|Name |Call |Default |Speed |Paired |Logged |Normalization |Distribution |Test |Function | |:-----------------------------------------------------------------------------------------------------------------------------|:------------|:-------|:------|:------|:------|:-------------|:---------------------|:-------------------------------------------|:------------------------| |StudentsT |DeT |Yes |Fast |Yes |Yes |No |Normal |Student's t test, equal variance |TDist {stats} | |limma |DeLimma |Yes |Fast |Yes |Yes |No |Normal |Empirical Bayes moderation |ebayes {limma} | |edgeR |DeEdgeR |Yes |Fast |Yes |No |Yes |Negative binomial |Exact/Likelihood ratio |exactTest {edgeR} | |DESeq2 |DeDeSeq |Yes |Fast |Yes |No |Yes |Negative binomial |Generalized linear model |DESeq {DESeq2} | |ABSSeq |DeAbsSeq |No |Fast |Yes |No |Yes |Negative binomial |Sum of counts |callDEs {ABSSeq} | |BGmix |DeBGmix |No |Fast |Yes |Yes |No |Normal |Bayesian mixture model |BGmix {BGmix} | |PoissonSeq |DePoissonSeq |No |Fast |Yes |No |Yes |Poisson log-linear |Poisson goodness-of-fit |PS.Main {PoissonSeq} | |RBM |DeRBM |No |Fast |No |Yes |No |Normal |Empirical Bayes & resampling |RBM_T {RBM} | |voom |DeVoomLimma |No |Fast |Yes |No |Yes |Log-normal |Empirical Bayes moderation |voom {limma} | |WelchsT |DeWelch |No |Fast |Yes |Yes |No |Normal |Welch's t test, unequal variance |TDist {stats} | |DEGseq |DeDegSeq |No |Medium |No |No |No |Binomial/Poisson |Likelihood Ratio Test |DEGexp {DEGseq} | |EBSeq |DeEbSeq |No |Medium |No |No |Yes |Negative Binomial |Empirical Bayesian |EBTest {EBSeq} | |NOISeq |DeNoiSeq |No |Medium |No |No |Yes |Nonparametric |Empirical Bayes |noiseqbio {NOISeq} | |PLGEM |DePlgem |No |Medium |Yes |No |No |Normal |Power Law Global Error Model |run.plgem {plgem} | |RankProd |DeRankP |No |Medium |Yes |Yes |No |Nonparametric |Rank product |RP {RankProd} | |SAM |DeSam |No |Medium |Yes |Yes |No |Normal |Alternative t test with permutation |samr {samr} | |SAMSeq |DeSamSeq |No |Medium |Yes |No |No |Nonparametric |Wilcoxon with resampling |SAMseq {samr} | |sSeq |DeSSeq |No |Medium |No |No |No |Negative Binomial |Shrinkage Approach of Dispersion Estimation |nbTestSH {sSeq} | |Wilcoxon |DeWilcoxon |No |Medium |Yes |Yes |No |Nonparametric |Wilcoxon signed-rank test |wilcox.test {stats} | |baySeq |DeBaySeq |No |Slow |Yes |No |No |Negative binomial |Empirical Bayesian |getLikelihoods {baySeq} | |bridge |DeBridge |No |Slow |No |Yes |No |T/Gaussian |Bayesian hierarchical model |bridge.2samples {bridge} | |LMGene |DeLMGene |No |Slow |Yes |No |No |Normal |Linear model & glog transformation |genediff {LMGene} | |ALDEx2 |DeAldex2 |No |Slower |Yes |No |Yes |Dirichlet |Welch's t/Wilcoxon/Kruskal Wallace |aldex {ALDEx2} | |BADER |DeBader |No |Slower |No |No |Yes |Overdispersed poisson |Bayesian |BADER {BADER} | |edgeRun |DeEdgeRun |No |Slower |Yes |No |Yes |Negative binomial |Exact unconditional |UCexactTest {edgeRun} | |tweeDEseq |DeTweeDeSeq |No |Slower |No |No |Yes |Poisson-Tweedie |Poisson-like |tweeDE {tweeDEseq} |